Construction Formula for Biological Age Estimation using Principal Component Analysis in Imbalanced Dataset

Aging is a natural process that leads to a decline in an individual’s physical health condition. While chronological age (CA) only measures the length of time someone has lived since birth, biological age (BA) offers valuable insight into an individual’s aging process. Consequently, BA proves to be a more effective indicator for evaluating the aging process than CA. Several methods have been developed to estimate BA. In this study, we employed Principal Component Analysis (PCA) to find important patterns of biomarkers in a dataset comprising 1245 individuals, specifically focusing on those within the age range of 30-65 and divided by gender. This study aims to utilized PCA for estimating the BA of the Indonesian population. PCA also allows for the assessment of the extent to which each biomarker contributes to the aging process. To address imbalanced dataset, the Synthetic Minority Oversampling Technique (SMOTE) was utilized. Biomarkers that truly influence BA will be selected using PCA. The relationship between CA and CBA shows at the value of coefficient of determination (r²) values, which were 0.51 for men and 0.46 for women. Furthermore, the fact that the mean of Corrected BA matches the mean of CA supports the model’s precision in estimating biological age.